Bent Nielsen
(1998)
'Conditional test for rank in bivariate canonical correlation
analysis'
ABSTRACT:
The likelihood ratio test for the hypothesis that the smallest
of two canonical correlations is zero is non-similar. This means
that the distribution of the test statistic depends on a nuisance
parameter: the value of the largest canonical correlation. If
this parameter were known it would be possible to construct a
test with a desired size. However, in applications the nuisance
parameter is usually estimated. It is therefore of interest to
evaluate the distributional properties of such a test,
conditional on the estimator. This distribution is found.
Although, it depends on the nuisance parameter, the dependency
seems to be negligible for practical purposes.